There are many things that can sabotage microwave QSO attempt. By the very nature of microwave antennas they have a narrow beamwidth which makes pointing the antenna accurately imperative. Various techniques have been suggested to minimize this error, the latest is the appearance of affordable digital compasses with onboard LCD display (1)

The other main cause of not making the QSO is not being on the correct frequency. Of course under very good conditions one can tune around and find the signal but when pushing the limits of a path and expecting a weak signal not looking on the correct frequency dooms the QSO to failure.

With improvements in the stability of local oscillators in the last 10 years the problem has lessened on the lower microwave bands (up to 10GHz). However with amateurs constantly pushing upwards in frequency to 24 and 47Ghz oscillator the problem has started to reappear.  For example on 47GHz where crystals can be multiplied typically 480 times to generate the local oscillator a 100Hz change of frequency (0.1ppm) of the crystal equates to a change in LO of 50kHz. The ideal solution would be to lock the local oscillator to a very stable low frequency reference oscillator and rely on the frequency accuracy of the I.F. radio. Alternatively generate an accurate signal in the desired band and use this as a known reference.

It is possible to obtain low frequency reference oscillators that maintain good accuracy and lock microwave oscillators to them. Unfortunately they tend to be expensive and hard to use in the field. One example is the Efratom FRK-1 Rubidium oscillator which is rated at 5x10^-11 accuracy but currently costs around $1000 on the surplus markets.

Fortunately the US government has a number of super accurate Caesium standards continually flying overhead in the Global Positioning System satellite constellation. They have to be very frequency accurate as the positioning system relies of very accurate timing. Any variation due to changing atmospheric conditions can be minimised by filtering and averaging. It is these GPS signals that are the basis for the equipment described here to accurately control the frequency of a 10MHz crystal oscillator, a microwave signal is then locked to this oscillator

The basic design was described by Shera in QST (2). His web site (3) contains further notes including vital corrections (4), without which notes, especially if hand wiring the unit you will not be able to make it work. A block diagram of the design is shown in Figure 1. The one pulse per second signal available from the GPS receiver is phase compared with the 10 MHz reference by using a gated oscillator and a counter. An error voltage generated by this comparison is used to fine tune the 10MHz reference frequency. The comparison and filtering is done digitally in a PIC microcontroller. This allows some intelligence to be added, for instance if the new correction value is wildly different from the previous one (as could be caused by a "spike") it is ignored.

Project Realisation

The first task was to find a cheap small GPS receiver that had a 1PPS output. Initial experiments were done with a Trimble Scoutmaster portable GPS unit, but this was wasted in this role and was better suited to navigation in the field. It was noted in the Shera notes that Motorola UT+ unit was available at a good price from (5) so one of these was purchased, complete with battery backup option to retain the settings with no power applied..

The harder task was to find a 10MHz reference oscillator with electronic Fine tuning (EFT). Shera recommended the Hewlett Packard HP10544A oscillator. Numerous sources were tried but the answer was always "Till the Shera article appears you couldn’t sell those oscillators for $25 in a frequency counter, now they are all gone" The only source found wanted $350 each, much too high, so the project was shelved for a while. This is where I hit lucky. Dave Clingerman W6OAL invited myself and Meg to stay with him for a few days after Microwave update 1998. Whilst touring his basement the latter version HP10544C oscillator was spotted. Dave said if I could use it I could have it, so it was traded for a PIC beacon keyer.

The main PCB was obtained from A&A engineering (6) and the other components were obtained from the sources quoted by Shera. An available LED display was used to display the control voltage rather than the LCD originally specified. The Hex code was downloaded from the Shera web page and programmed into a 16C73/JW. The reprogrammable EPROM version of the chip rather than the cheaper One Time Programmable PIC16C73A-20 was used to allow for future software upgrades. Programming was achieved using Microchip's MPLAB-IDE program available free from Microchip's web page (7) and a PICSTART PLUS Programmer which I obtained from Digikey (8). A PCB for the PSU components was designed to run the unit off 110 or 220V AC and provide 12V for the 10MHz oscillator, 24V for its heater, -5V for the main PCB and 5V for the GPS receiver and LED Display.

Two 9 pin RS232 connectors were mounted on the back panel. One was used to allow communication to and from the GPS unit, spare sections of IC3 were used for level conversion. Note that the UT+ can only be communicated with and can only produce Motorola Binary protocol signals, (rather than the normal NMEA strings). For a price software can be obtained from Motorola to talk in this protocol but a cheaper method is to obtain the TAC32 software from (9) which has the binary protocol ability to communicate with the unit for such purposes as putting it into "position hold mode". The other connector is used to monitor the diagnostic information being provided from the PIC Microcontroller, which can also be used to graph the performance/stability of the system. Details of the interconnections between the constituent parts of the GPS locked reference are shown in Table 1
 

Having built the first unit it proved to be so useful it was decided to build a second one. This would allow the frequency to be set accurately at both ends of a path. The only difference between the units is that the second one uses a Motorola UT GPS unit obtained surplus at Dayton 1999 for $75 and a 10MHz source made by in the UK by HCD Research which runs off a single 12V supply. The following paragraphs describe some of the uses the 10MHz GPS locked frequency standard has been put to:-

1. 10GHz frequency stable beacon

The first use for the unit was to make an accurate beacon for 10368MHz. This was easy to implement as WB6IGP was distributing a Qualcom 12GHz PLL board which accepted a 10MHz reference and could be programmed to produce a 10mW 2592MHz signal by choosing M=63, R=4 and A=8. This signal is used to drive a G3WDG001 times four multiplier to produce 10368MHz at +17dBm (10). Further details of the synthesiser board are contained elsewhere in the proceedings and also in (11).



In the original application the synthesisers shifted frequency between the receive and transmit. The switching circuitry in the "Texas" style version of the synthesiser has successfully been used to switch the synthesiser between 2592 and 2556MHz which is the drive frequency for 10224MHz used to convert 10368 to 144MHz (for 2556MHz M=62, R=4 and A=9)

Having gone to all the trouble of maintaining frequency accuracy of the beacon it was pointless to corrupt this accuracy by using FSK to identify the signal, so the oscillator unit is fed through an Isolator to a pin diode switch which amplitude modulates the signal with a callsign. Keying was provided by a WW2R PIC keyer as described in (12) .The block diagram of the beacon is shown in Figure 2.

Luckily for those in the USA the "middle" microwave bands are still harmonically related (unlike Europe where they are anything but!). If an 1152MHz source locked to the 10MHz reference could be made, markers on 2304.000, 3456.000 and 5760.000 and 10368.000 could be generated. Fortunately WB6IGP also had another Qualcom PLL board that could be programmed for 1152MHz, further details of which is contained elsewhere in the proceedings. This runs off 5V and 15V, produces 10mW and provides good harmonics through 10GHz without external circuitry. When driving the circuitry by G3PYB described in (13) using an SMA socket, LNC diode and DC return wire, a healthy signal can also be heard on 24192MHz

It is not possible to use the same synthesiser as on 10GHz for this band, as to generate a signal using currently available methods on 24192MHz requires drive on 2419.2MHz and the synthesiser can only be programmed in 2MHz steps. A different method has to be employed.

Looking back in microwave update proceedings the design to lock a crystal around 100MHz to a 10MHz reference by WA6CGR was rediscovered (14). Also found was the web page of VK5KK (15) which was similar in concept to the WA6CGR which whilst lacking it's versatility was much smaller. PCBs are available from VK5KK for the design.

A G4DDK004 oscillator unit (16) was chosen to generate the 2419.2MHz 10mW signal starting with a 100.8MHz crystal cut for 50 degrees centigrade operation with a corresponding temperature posistor mounted on it. A sample of the 100.8MHz was extracted from the base of TR1 via a 1.8pF capacitor and amplified in an MSA06 modamp to a level of +5dBm and fed into the VK5KK board along with the 10MHz GPS derived reference. The VK5KK board is set with M=50 and N=4 with L4 employing 42 turns. The control voltage produced by the board is applied to a BB105 varicap diode connected across L1, completing the control loop. The oscillator unit is followed by a 2GHz isolator and pin diode switch to provide Ampitude modulation identification) then a G3WDG011 times five multiplier and a DB6NT doubler using a pair of MGF1302 which produces 10mW on 24GHz. A block diagram of the whole 24GHz "mini beacon" is shown in Figure 3.

By using a 98.1MHz crystal and tuning the G4DDK004 to 2354.4MHz a signal can be generated on 23544MHz using the above lineup then driving a diode doubler to 47088.0MHz allowing calibration on that band.

Appearing on the surplus market recently have been a number of Adret type 5104 synthesisers which synthesise a 10dBm signal in the range 90 to 119.99999MHz in 10Hz steps. They were being used to drive microwave PLL bricks in satellite earth stations their frequency agility allowing a rapid QSY. They have an accuracy of 2 parts in 10^-8 per day after 72 hours operation when using their internal reference but this accuracy can be increased dramatically by external locking to the GPS reference. Further details of their usage, including interfacing details can be found on G4DDK's web page (17), should any appear in the USA.

Hopefully the above description will help remove the uncertainty in knowing ones microwave frequency enabling more contacts over more marginal paths to be completed.

1. Jameco type NF200 part number 149921 (www.jameco.com)
2. A GPS-Based Frequency Standard, B. Shera, QST July 1998 p. 37-44.
3. http://ww.rt66.com/~shera
4. http://www.rt66.com/~shera/Q&A.html
5. http://ww.synergy-gps.com
6. A&A Engineering, 2521 W La Palms, Unit K, Anaheim, CA92801
7. http://www.microchip.com
8. http://www.digikey.com
9. http://www.cnssys.com/cnsclock/Tac32Software.html
10. The G3WDG-001 2.5 to 10GHz multiplier/amplifier. C.Suckling, G3WDG RSGB Microwave Handbook, Volume 3 pp18.100-18.107
11. Programming surplus synthesizers on 10GHz. B.Wood N2LIV & Bob Schoenfield WA2AQQ Miccrowave update proceedings 94 PP275-279
12. The WW2R Beacon keyer, D.Robinson WW2R. Microwave update proceedings 98 pp97-105. http://www.flash.net/~g4fre/compproj.htm
13. http://www.geocities.com/SiliconValley/Vista/7012/marker.html
14. A Universal phase lock loop system for microwave use. D Glawson WA6CGR Microwave update 94 proceedings pp69-76.
15. PLL for microwave local oscillators http://www.ozemail.com.au/~tecknolt/Projects/vk5kk48.htm
16. An Oscillator source for 2.0 to 2.6GHz. S.Jewell, G4DDK RSGB Microwave Handbook Volume 2 pp8.12 -8.16 The full range of G4DDK oscillators is described at http://www.btinternet.com/~jewell/DDK_Des.htm
17. http://www.btinternet.com/~jewell/Adret.htm